A diamond has the highest thermal conductivity among a variety of materials, and has the highest breakdown electric field strength among semiconductors. Accordingly, the diamond is the semiconductor material most suitable for a high-power semiconductor requiring high voltage, large current operation. Since holes and electrons in the diamond have a high mobility and saturated velocity, the diamond is suitable for a high-frequency semiconductor element operable in high frequencies. A high-frequency diamond semiconductor element is a semiconductor element controlling high power in a high frequency band including a micro-wave band region and a millimeter-wave band region.
FIGS. 5A to 5C are diagrams illustrating conventional process-steps in producing a diamond semiconductor element. A description will be given below of the process-steps in producing a diamond semiconductor transistor by the use of a conventional technique disclosed in non-patent document 1 (see non-patent document 2 regarding surface orientation).
First, a single-crystal diamond substrate 1-31 is prepared as shown in FIG. 5A. The surface orientation of the surface of the single-crystal diamond substrate 1-31, which is{right arrow over (s)}  [Expression 1]has its surface orientation preciously in the [001] direction.
Then, as illustrated in FIG. 5B, crystal growth is performed on the single-crystal diamond substrate 1-31 to form a single-crystal diamond thin-film 1-32 thereon. In the crystal growth process-step, a two-dimensional-form hole channel 1-33 is formed in such a manner as to be parallel to the surface of the single-crystal diamond thin film 1-32. The surface orientation of the surface of the single-crystal diamond thin-film 1-32, which is{right arrow over (d)}  [Expression 2]and, the surface orientation of the forming face of the hole channel 1-33, which is{right arrow over (c)}  [Expression 3]becomes equal to the surface orientation of the surface of the single-crystal diamond substrate 1-31, and thus has the surface orientation preciously in the [001] direction.
Then, as illustrated in FIG. 5C, a source electrode 1-34, a gate electrode 1-35, and a drain electrode 1-36 are all formed on the single-crystal diamond thin-film 1-32. The longitudinal direction of the gate electrode 1-35, which is{right arrow over (g)}  [Expression 4]is the [100] direction.
The characteristics of the transistor produced by the aforementioned conventional method are disclosed in detail in non-patent document 1. The transistor characteristics data on all items with reference to the characteristics of a transistor having a gate length of 0.2 μm (normalized by the gate width) is disclosed. According to non-patent document 1, the maximum transconductance gmmax of the transistor is 150 mS/mm at the most.
However, a diamond single-crystal has the problem of an extremely high density of a crystal defect as compared with that of another semiconductor, for example, silicone, gallium arsenide, indium phosphide, gallium nitride, and the like. For this reason, the natural physical properties of a diamond, such as high thermal conductivity, a high breakdown electrical field, satisfactory high-frequency characteristics and the like, cannot be reflected in the transistor characteristics. A transistor using a diamond semiconductor is not yet in actual use. This problem is described in non-patent document 1. In consequence, for the achievement of the practically useful transistor using a diamond single-crystal, what is required is to produce an element by a method that minimizes the occurrence of the crystal defect peculiar to a diamond.
It is a first object of the present invention to significantly suppress occurrence of a crystal defect specific to a diamond by shifting the surface orientation of a diamond substrate slightly from the [001] direction.
It is theoretically clear that because a diamond has the highest thermal conductivity among materials and has the highest breakdown electric field strength among semiconductors, the diamond is the semiconductor material most suitable for a high-power semiconductor operable at high voltage and large current. In addition, it is known that because a diamond has a high mobility and saturated velocity of electrons and holes, the diamond is suitable for a high-frequency semiconductor element operable at high frequency.
FIGS. 13A to 13G illustrate process-steps in producing a conventional diamond semiconductor element. Gold (Au) is evaporated onto a diamond single-crystal thin-film 2-11 (FIG. 13A) having a two-dimensional hole channel close to the surface to form an Au thin-film 2-12 (FIG. 13B). The Au thin-film 2-12 is coated with a resist 2-13 (FIG. 13C). Then, photolithography or an electron beam is applied for exposure and development to remove part of the resist 2-13 to form an aperture in the resist 2-13 above the area in which a gate electrode will be formed (FIG. 13D). Then, a sample is immersed in an Au etchant to etch a portion of the Au thin-film 2-12 close to the aperture in the resist 2-13 (FIG. 13 E).
As illustrated in FIG. 13E, the etching is performed on the portion of the surface of the Au thin-film 2-12 exposed by the aperture of the resist 2-13, then proceeds from there in the depth direction (the direction at right angles to the diamond single-crystal thin-film 2-11) and simultaneously in the lateral direction (the direction horizontal to the diamond single-crystal thin-film 2-11). For this reason, an area of the Au thin-film 2-12 below the resist 2-13 is also cut away. The portion cut away or, as it were, hollowed out is called the “undercut”. When the Au thin-film 2-12 is etched in the lateral direction by the etchant in this manner, because the bonding strength between the Au thin-film 2-12 and the resist 2-13 is greater than the bonding strength between the Au thin-film 2-12 and the diamond single-crystal thin-film 2-11, the etching velocity in the lateral diction is slower close to the resist 2-13 and faster close to the diamond single-crystal thin-film 2-11. For this reason, an angle θ of the end face of the etched portion of the Au thin-film 2-12 is about 45 degrees. In other words, each section of the Au thin-film 2-12 divided into two by the etching has an inverted mesa shape having the upper side wider than the lower side.
Next, Al (aluminum) is evaporated (FIG. 13F). The Al, which passes through the apertures of the resist 2-13 and the Au thin-film 2-12 to be evaporated directly onto the surface of the diamond single-crystal thin-film 2-11, and the Al, which is evaporated onto the resist 2-13, respectively form Al thin-films 2-15G, 2-15. Then, a sample is immersed in a liftoff fluid for the liftoff of the resist 2-13 to remove the resist 2-13 and the Al thin-film 2-15 evaporated thereon (FIG. 13G). At this stage, one of the Au thin-films 2-12 is defined as a source electrode 2-16S and the other thin-film 2-12 is defined as a drain electrode 2-16D, and also, the Al thin-film 2-15G remaining on the diamond single-crystal thin-film 2-11 is defined as a gate electrode 2-17G. At this point, the thickness tS, tD is 0.6 μm, and the gate length dG corresponding to the length from the end of the gate electrode 2-17G close to the source to the end close to the drain is 0.2 μm.
The diamond semiconductor has the physically derived limitation that the channels in which the electron and the hole of a transistor travel are required to be located within 0.1 μm from the surface unlike other semiconductors, for example, silicone, gallium arsenide, indium, gallium nitride, and the like (see non-patent document 3).
Under this theoretically limitation, a requirement for an increase in the transconductance gm, which is the degree of amplification of the diamond semiconductor element, to produce an enhancement up to a practical level in the maximum oscillation frequency fmax, which is the upper limit to operation frequency in the high-frequency characteristics, is to reduce the source-gate electrode distance dSG between an end of the face of the source electrode 2-16S making contact with the diamond single-crystal thin-film 2-11 and the end of the gate electrode 2-17G close to the source, and the gate-drain electrode distance dGD between the end of the gate electrode 2-17G close to the drain and an end of the face of the drain electrode 2-16D making contact with the diamond single-crystal thin-film 2-11. This presents fewer problems in the cases of other semiconductors, but in the diamond semiconductor element this is a critical problem resulting from its physical properties that needs to be solved. In addition, a reduction as great as possible in the gate length dG in FIG. 13G is also required for an increase in the maximum oscillation frequency fmax.
Non-patent document 4 discloses that the source-gate electrode distance dSG and the gate-drain electrode distance dGD are respectively from 1.3 μm to 1.4 μm, because the distance between the source electrode 2-16S and the drain electrode 2-16D is 2.6 μm or 2.7 μm, and the gate length dG is 0.2 μm.
At the same time, for the prevention of an unwanted voltage drop from occurring in transistor operation, the source electrode resistance and the drain electrode resistance are required to be reduced as much as possible. For a reduction in the source electrode resistance and the drain electrode resistance, the thickness tS of the source electrode 2-16S and the thickness tD of the drain electrode 2-16D are required to be increased as much as possible.
However, in the process of etching the Au thin-film 2-12, the inverted mesa structure having an angle θ of about 45 degrees, as shown in FIG. 13E, occurs at the end faces of the source electrode 2-16S and the drain electrode 2-16D close to the gate electrode 17G. This gives rise to the problem of the impossibility of reducing the source-gate electrode distance dSG and the gate-drain electrode distance dGD so as to be smaller than the thickness tS of the source electrode 2-16S and the thickness tD of the drain electrode 2-16D. In other words, the conventional art is incapable of simultaneously satisfying the two requirements, “a reduction in the source-gate electrode distance and the gate-drain electrode distance” and “an increase in the thickness of the source electrode and the thickness of the drain electrode”.
FIGS. 14A to 14C show the characteristics of a diamond transistor produced by a conventional process. They are the results disclosed in non-patent document 3, all of which are standardized with reference to the characteristics of a transistor having a gate length of 0.2 μm. In the drain current-voltage characteristics shown in FIG. 14A, the maximum drain current normalized by the gate length dG is 0.35 A/mm at the most. In the dependence of the transconductance gm on the gate voltage VG (transfer characteristics) shown in FIG. 14B, the maximum transconductance gmmax normalized by the gage length dG is 150 mS/mm at the most. Further, in the dependence of the power gain U on the frequency f shown in FIG. 14C, the maximum oscillation frequency fmax which is an upper limit to the operation frequency is 81 GHz at the most. The drain breakdown voltage, which is not shown in figures, is 45V at the most.
Accordingly, it is a second object of the present invention to achieve the compatibility between “a reduction in the source-gate electrode distance dSG and the gate-drain electrode distance dGD” and “an increase in the thickness tS of the source electrode and the thickness tD of the drain electrode” to increase the maximum oscillation frequency fmax for a great improvement in the characteristics of a diamond field-effect transistor, and also to bring the voltage drop down, thus reaching a practical level.
Next, a process for producing a conventional diamond single-crystal thin-film will be described using FIGS. 22A and 22B. A diamond single-crystal substrate 3-1 having surface orientation (100) is prepared (FIG. 22A). Then, a microwave plasma CVD (chemical vapor deposition) apparatus is used to deposit a diamond single-crystal thin-film 3-2 of about 1 μm to 5 μm thickness onto the diamond single-crystal substrate 3-1 at a substrate temperature of 700° C. using methane as a reaction gas (FIG. 22B). The surface of the diamond thin-film obtained by the CVD technique is hydrogen-terminated in an as-grown state and has surface conduction properties, and functions as a P-type semiconductor.
Non-patent document 5 describes that, in order to improve the crystallinity of a diamond thin-film deposited on a silicone substrate, the above diamond thin-film is placed in a ceramic tube in which a vacuum of 1×10−6 Torr is produced, and high-temperature annealing of 1000° C. or more is performed in the vacuum.
Then, the diamond single-crystal thin-film produced by the process for producing the conventional diamond single-crystal thin-film described in FIGS. 22A and 22B has an average mobility of about 800 cm2/Vs at room temperature, and a high quality thin-film is obtained with satisfactory reproducibility. However, impurities and a large number of crystal defects such as a growth hillock, abnormal growth particles and the like exist in the above diamond single-crystal thin-film.
In non-patent document 5, when the temperature at which the annealing is performed reaches 1200° C. or more, band-A emissions associated with deterioration of the diamond thin-film (emission resulting from defects) are increased. In short, in patent document 2, deterioration in crystallinity is increased at 1200° C. or more.
Because the higher the annealing temperature becomes, the more the crystallinity is improved, the annealing is desirably performed at higher temperatures. However, in non-patent document 5, if the temperature is increased for an improvement in crystallinity, when the temperature reaches a certain degree (1200° C.) or more, the conversion to graphite progresses in the diamond thin-film, resulting in an increase in deterioration in crystallinity.
Accordingly, it is a third object of the present invention to provide a diamond-thin-film producing process which is capable of reducing crystal defects, impurities and the like existing in a diamond thin-film to produce a high quality diamond thin-film.
It is known that a diamond has physical characteristics as a semiconductor superior to those of silicon (Si). It is recognized in a theory that the diamond element has characteristics regarding high-temperature operation five times that of the Si element, high-voltage performance 30 times that of the Si element and an increase in speed three times that of the Si element. For this reason, it is expected that the diamond will realize a high-output device having a high thermal conductivity and a breakdown electric field strength, a high-frequency device having a high carrier mobility and a high saturated drift velocity, and the like. In other words, because a field-effect transistor (FET) or a bipolar transistor uses a diamond semiconductor, an electron element capable of driving at a high frequency for high-power operation, significantly exceeding the conventional semiconductors, is provide. In addition, it is clear in theory that the realization of a semiconductor laser and a light emitting diode using a diamond semiconductor realizes a high intensity light emitting element with a wavelength of 225 nm in the ultraviolet region (see non-patent document 6).
A diamond has a band gap of 5.5 eV and is originally an insulator, but, as in the case of Si, if the diamond is doped with B which is a III group element, an acceptor level occurs, resulting logically in a p-type semiconductor.
Since a p-type semiconductor layer in a transistor or optical device structure has a high resistance, as in the case of an insulator when the hole concentration is less than 1.0×1015 cm−3, the p-type semiconductor layer does not fully function as a p-type semiconductor, and is thus of no use. Also, since a p-type semiconductor layer in a transistor or optical device structure shows a metallic electrical conduction when the density of the dopant element exceeds 1.0×1021 cm−3, in this case, the p-type semiconductor layer also does not fully functions as a p-type semiconductor, and is thus of no use. Thus, in the p-type semiconductor layer the hole concentration is required to be 1.0×1015 cm−3 or more and the dopant element concentration to be 1.0×1021 cm−3 or less.
Also, because the hole concentration and the dopant element concentration in a semiconductor depend on temperatures, it is important to satisfy the above requirements in operational temperatures around room temperature (300K) in order to ensure the practicality of the device. In addition, in high power use such as in appliances, electrical apparatuses, industrial equipment and the like, operation in a high-temperature condition is particularly required. For this reason, there is a necessity to satisfy the above requirements in, for example, about 500K which is higher than room temperature.
However, as shown in FIG. 38, the conventional technique of doping a diamond with boron (B) has the problem of only a hole concentration of 6×1014 cm−3 being obtained at room temperature (300 k) even when the B-atom concentration is 1.0×1021 cm−3. In FIG. 38, the horizontal axis represents the measurement temperature (K) and the vertical axis represents the hole concentration (cm−3) in a conventional p-type diamond semiconductor, in which the measured values at each B-atom concentration in the conventional p-type diamond semiconductor are plotted. These values do not satisfy the 1.0×1015 cm−3 required for practical use at 300K. As a result, there arises the problem of the incapability of the practical use of a diamond semiconductor as a transistor or an optical device.
As means for increasing the hole concentration at about 300K of the diamond semiconductor, a greater increase in the B-atom concentration than 1.0×1021 cm−3 is conceivable. However, the crystal quality of a diamond becomes poor as the B-atom concentration becomes greater than 1.0×1021 cm−3, so that the diamond loses its semiconductor properties, resulting in the problem that it is of no practical use.
Accordingly, it is a fourth object of the present invention to provide a practically useful p-type diamond semiconductor having a hole concentration of 1.0×1015 cm−3 or more at room temperature (300K) or more and a dopant atom concentration of 1.0×1021 cm−3 or less, and to provide a process for producing the same.
A diamond is a semiconductor having both the highest thermal conductivity (22 W/cmK) and the highest breakdown electric field (>10 MV/cm) among substances as described above, and also a high carrier mobility (electron: 4500 cm2/Vs, hole: 3800 cm2/Vs), and if highly efficient doping is accomplished, a transistor operating at a high frequency and high output surpassing that of Si, GaAs, GaN is realized.
One of the methods of doping a diamond is an ion implantation technique. The ion implantation technique is a method for accelerating impurities at high voltage so as to lend them energy of some kV to some MV for the introduction of impurity ions into the crystal, in which because a high energy process is involved, damage (crystal defects, amorphous layer and the like) occurs in the crystal in proportion to the acceleration voltage. The damage can be removed by performing appropriate high-temperature annealing treatment, with the result that the dopant is electrically activated and the semiconductor characteristics caused by the implanted impurities emerge. However, since a diamond is thermodynamically a meta-stable layer under ordinary pressure (1 atmospheric pressure), a high-quality diamond semiconductor is not obtained by the generally employed annealing process performed under ordinary pressure or vacuum. For this reason, in patent document 1 high-temperature annealing under high pressure is performed as described below.
FIGS. 43A to 43E illustrate the process-steps in producing a diamond semiconductor using an ion implantation technique according to conventional art. An ion implantation apparatus is used to implant dopant (boron) into a diamond single-crystal (FIG. 43A) under the conditions of an acceleration voltage of 150 kV and a dose of 1×1016 cm−2 (FIGS. 43B to 43C) and then firing (annealing) is performed for one hour at a pressure and a temperature of 5 GPa and 1700K (FIG. 43D).
However, the diamond thin-film subjected to the annealing treatment in this manner has a high resistance, thus giving rise to the problem of exhibiting no semiconductor characteristics. This is because etching occurs on the diamond surface during the process of the high-temperature and high-pressure annealing treatment so as to cut away the layer into which the ion is implanted (FIG. 43D). In this manner, the conventional technique has the problem of the impossibility of providing a diamond semiconductor because the diamond layer formed by ion implantation is etched during the high-temperature and high-pressure annealing.
Accordingly, it is a fifth object of the present invention to provide a process for producing a diamond semiconductor which is capable of preventing the diamond surface from being etched by the high-temperature and high-pressure annealing performed on ion-implanted diamond to provide high-quality P-type and N-type diamond semiconductors which cannot be provided by conventional processes.    Patent document 1: Japanese Patent Publication No. 8-15162    Non-patent document 1: Makoto KASU et al., “High-frequency characteristics of diamond MESFET”, Journal of the Japan society of applied physics “Applied Physics”, vol. 73, No. 3 (March, 2004), pp. 363-367    Non-patent document 2: C. Kittel, “Introduction to Solid State Physics”, published by Maruzen, the fifth edition, first volume, pp. 11-22    Non-patent document 3: Makoto KASU and 6 others, “High-frequency characteristics of diamond MESFET”, Journal of Applied Physics, the Japan society of applied physics, 2004, vol. 73, No. 3, pp. 0363-0367    Non-patent document 4: M. Kasu, “Influence of epitaxy on hydrogen-passivated diamond” Diamond and Related Materials, 2004, No. 13, pp. 226-232    Non-patent document 5: J. Ruan et al., “Cathodoluminescence and annealing study of plasma—deposited polycrystalline diamond films” J. Appl. Phys. 69(9), 1 May 1991    Non-patent document 6: Kasu et al., “High-frequency characteristics of diamond MESFET”, Journal of Applied Physics, 2004, vol. 73, No. 3, pp. 363-367    Non-patent document 7: F. P. Bundy, H. P. Bovenkerk, H. M. Strong, and R. H. Wentorf, “Diamond-Graphite Equilibrium Line from Growth and Graphitization of Diamond”, The Journal of Chemical Physics, August, 1961, Vol. 35m Number 2, pp. 383